Audio-visual segmentation for content-based retrieval
David Pye, Nicholas J. Hollinghurst, Timothy J. Mills, Kenneth R. Wood · 1998
This paper reports recent work at ORL on segmentation of digital audio/video recordings. Firstly, we describe an audio segmentation algorithm that partitions a soundtrack into manageably sized segments for speech recognition. Secondly, we present an algorithm for detecting camera shot-break locations in the video. The output of these two algorithms is combined to produce a semantically meaningful segmentation of audio/video content, appropriate for information retrieval. We report the success of the algorithms in the context of television news retrieval.